JPS6273299A - Voice recognition system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] (Industrial application field) The present invention relates to a speech recognition method with high recognition accuracy.

(Prior Art) Research and development regarding speech recognition has been carried out for the purpose of increasing the efficiency of inputting information and communication equipment and improving system functions. A common method for performing this speech recognition is the pattern matching method.

First, prior to explaining the present invention, a conventional pattern matching method will be explained with reference to FIG.

In FIG. 6, lO is a voice input terminal, 11 is a voice analysis section, 12 is a section detection section, 13 is an input memory section, 14 is a comparison pattern memory section, 15 is a similarity calculation section, 16 is a judgment section, and 17 is a It is an output terminal.

In this conventional recognition method, input speech inputted to the speech input terminal 10 is processed in the speech analysis section 11 to generate a time-series pattern of vectors representing characteristics (hereinafter referred to as "speech pattern").
Convert to This audio pattern is generally created by sampling (hereinafter referred to as sampling) in-band frequency components extracted by two band-pass filter groups with different center frequencies at every time interval TO (for example, 8 milliseconds). obtained by. On the other hand, the voice analysis unit 11 calculates one voice at a time point corresponding to the voice pattern. The voice patterns calculated in the voice analysis section 11 are sequentially stored in the input memory section 13, and the voice power is outputted to the section detection section 12.

The section detection section 12 determines a speech section, that is, the start and end of the speech, based on the speech power from the speech analysis section 11. The algorithm for determining the start and end of the voice based on the voice power is a complex algorithm as disclosed in Japanese Patent Application No. 108668/1982, and the time when the voice power exceeds the threshold is determined as the start and end of the voice. There are simple algorithms and other algorithms that consider the point at which the sound ends as the end of the voice, and any suitable algorithm is used to detect the section. The audio pattern between the start and end points determined by the section detection section 12 is read out from the input memory section and sent to the similarity calculation section 15. On the other hand, a comparison pattern is separately inputted to the similarity calculation unit 15 from the comparison pattern memory 14. This comparison pattern is a time-series pattern of vectors obtained by subjecting words to be recognized (hereinafter referred to as categories) to the same speech analysis process as speech patterns, and is stored in the comparison pattern memory section 14 in advance.

When storing this, a comparison pattern is created, but the creation method differs depending on the purpose of recognition. For example, in the case of a recognition method that limits speakers, the frequency analysis unit 11 A speech pattern obtained by using this method or by performing an equivalent speech analysis process is stored in the comparison pattern memory section 14 as a comparison pattern.

The similarity calculation unit 15 calculates the similarity between the speech pattern and the comparison pattern.
"Oki Electric Research and Development No. 118", 48, (3) (1982)
The weighting disclosed in December 2003, pp. 53-58) uses a method called a linear matching method or other suitable method.

Using the similarity for each category output from the similarity calculation section 15, the 1 determination section 16 outputs the category name given to the comparison pattern giving the maximum similarity as a recognition result.

The above is an outline of the conventional speech recognition method using the pattern matching method.

(Problems to be Solved by the Invention) The conventional recognition method described above evaluates the difference between a speech pattern that gives the shape of the speech spectrum and a comparison pattern calculated in advance by the same analysis process using a measure of similarity. , the recognition result was the category name of the comparison pattern that gave the greatest degree of similarity. Therefore, when the category of the speech pattern and the category of the comparison pattern are the same, the similarity is large, and when they are different, the similarity is small. However, the shape of the speech spectrum may be affected by factors other than speech, such as If distortion occurs due to external noise, the degree of similarity between the two cannot be said to be large, even if they are in the same category.

Further, in the conventional recognition method, the calculation process takes time and requires a large storage capacity, so there is a problem that the structure of the device implementing this method becomes large.

SUMMARY OF THE INVENTION In view of these conventional problems, it is an object of the present invention to provide a speech recognition method with good recognition accuracy even in a noisy environment.

Another object of the present invention is to provide a voice recognition system which, when configured as a device, has a compact and in-cylinder structure, has a high calculation processing speed, and requires a small storage capacity.

(Means for Solving the Problems) In order to achieve the above-mentioned object, the speech recognition system of the present invention takes the following measures.

(a) First, the frequency components of the input audio are extracted by multiple bandpass filters, and the output is
(referred to as audio frames) and calculates a feature vector.

(b) Also, calculate a noise pattern obtained by time-averaging feature vectors in a predetermined noise section that is known in advance to be only noise.

(C) After this noise pattern extraction, a speech feature vector is calculated by subtracting the noise pattern from the feature vector.

(d) A least squares approximation straight line is calculated for the audio frame A σ from the audio feature vector described above, and the local value obtained by setting the component corresponding to the channel that is maximum in the frequency axis direction to 1 with this least squares approximation straight line as a reference. Calculate the peak vector.

(e) Calculate the frame power in the audio frame from this audio feature vector, and calculate the start and end of this frame power.

(f) The local peak vector calculated for each audio frame from the start end to the end is linearly expanded or contracted on the time axis so that it becomes a constant audio frame length.

(g) During the registration process, for each voice of the recognition target word, a comparison pattern is created by performing processes corresponding to each of the above-mentioned processes (a) to (f) performed on the input voice (registration process ).

(h) The above (a) applies to the voice uttered after recognition processing.
Linear matching processing is performed between the input pattern obtained through the processing up to (f) and the comparison pattern to calculate the degree of pattern similarity between the comparison pattern and the input pattern.

(i) Processing is performed in which the category name added to the comparison pattern that provides the maximum similarity among the pattern similarities calculated for each comparison pattern is used as a recognition result.

In the manner described above, the result of recognizing input speech is obtained.

The above-mentioned processes (a), (b), and (c) are processes for extracting only speech from an input with high noise, and are considered difficult under high noise conditions (e) This is a process that facilitates the process of detecting the voice section of a term.

Furthermore, recognition performance in a high noise environment is improved by using the local peak vector calculated in section (d) to calculate the similarity between sections (h) and (i). This is because a local peak vector calculated based on a position giving a peak of the audio spectrum is used for similarity calculation, instead of using a vector giving the shape of the spectrum as in the conventional method. Therefore, it is based on the fact that when noise is mixed, the shape of the spectrum changes significantly, but the position of the peak of the spectrum does not change.

(effect) Next, the operation of this invention will be explained.

The functions for achieving the speech recognition method of the present invention are constituted by each processing section shown in FIG.

The detailed processing will be explained below.

The sound is converted into an electrical signal through a microphone, passed through an amplifier (not shown), a low-pass filter (not shown), and sent to an A/D converter (not shown), where it is sampled every 83 microseconds. ), then input to the input terminal 21.

Each of the above-mentioned items will be explained below.

[Feature vector calculation process in section (a)] Frequency analysis of audio data input to the input terminal 21 is performed by the feature vector calculation unit 22, and the data is converted into audio frame time series feature vectors.

The feature vector calculation unit 22 includes a plurality of band-pass filters each having characteristics with different center frequencies as shown in FIG. 2 for frequency analysis, a low-pass filter, and sampling for each audio frame. sampling means (each not shown in the figure).

Each band filter extracts only the center frequency component from the voice. The data series separated by each band filter in this way is called a channel. After performing an absolute value calculation on the bandpass output of each channel, the output is input to a low-pass filter. The low-pass filter output for each channel is resampled by a sampling means every audio frame period to obtain the components of the feature vector.

Now let the output of the low-pass filter of the channel in the i-th audio frame be al, then the feature vector ai in the i-th audio frame is a=(a
l,a! ',-,aL・=,a''IL l
It can be expressed as l. Here, K is the number of channels.

[Noise pattern calculation process in section (b)] This process is performed by the noise pattern calculation unit 23. A period in which only noise is input and no voice is input is set, for example, to 10 consecutive voice frames (the number of voice frames is not woody), and this is called a noise period.

The feature vector of a noise section represents the spectral shape of the noise, and is especially called a noise vector.The average value of the noise spectrum within the noise section is calculated by It is called a pattern.

Let the component of the noise pattern N be Nk.

Kano = (Nl, N2.... Hk,... HK
It becomes H.

[Section (C) Audio feature vector calculation process] This process is performed by the audio feature vector calculation unit 24.

After the noise interval, that is, after the noise pattern calculation, the noise pattern N from the noise pattern calculation unit 23 is subtracted from the feature vector a1 output from the feature vector calculation unit 22, and the voice feature vector w''''b> , b> + '''
+ b and ``' + ''' 1 is calculated by the following formula.

This processing in the processing unit 24 is a method for improving speech recognition performance in a high-noise environment, and is effective when noise continues relatively steadily.

[Local peak vector calculation process in section (d)] This process is performed by the local peak calculation unit 25.

The audio feature vector Toi sent from the audio feature vector calculator 24 is calculated by the local peak vector calculator 25.
Convert to local peak vector town at .

This conversion process will be explained with reference to FIGS. 3(A) to 3(C).

Each component of the audio feature vector Toj is logarithmically transformed using the following equation.

Figure 3 (A) shows the logarithmic transformation of this voice feature vector component.
1(k), where the channel number k is plotted on the horizontal axis and Xl (k) is plotted on the vertical axis. This figure represents the shape of the logarithmic spectrum of the audio in the i-th audio frame. has been done.

Next, the least squares approximation straight line can be learned by Eq. Perform normalization using .

Zt (k) = x, (k) −Yl (k)=
xi(k) −u,(k)−k −vi(k)
(51 This normalized speech feature vector component z1
An example of (k) is shown in FIG. 3(8), in which the channel number is plotted on the horizontal axis and z4 (k) is plotted on the vertical axis in FIG. 3(B).

Next, this zi(
k) to calculate the local peak vector Ei.

For k that satisfies the judgment condition of equation (6), r−ゝ=
1. For k that does not satisfy, r-=o, 1
.

Vector having values as components rI F1 = (r!r',・"+rto・"+411'
Calculate 1. This vector ri is called a local peak vector. This local peak vector ri
An example of this is shown in FIG. 3(C).

[Voice section detection processing in section (e)] This process is performed by the voice section detection section 26.

Using the audio feature vector calculated by the audio feature vector calculation unit 24 for each audio frame, the frame power Pi of the audio frame is calculated.

In the speech section detection unit 2B, the speech feature vector Ib
Voice section detection is performed using the frame power Pi obtained from i.

As mentioned above, various algorithms have been proposed for detecting speech intervals, but the purpose of this invention is not to use the algorithms themselves, but to detect speech intervals by subtracting the noise pattern N from the feature vector Therefore, for convenience of explanation, here, for convenience of explanation, the voice frame in which the frame power P is equal to or greater than the predetermined threshold Ps is defined as the voice start point I5, and the frame power P from the voice start point is , is the threshold Ps
The audio frame in which the number is less than 1 is considered to be the end of audio I6.

In Figures 4 (A) and (B), the input voice is "Sarapolo", and car noise is added as noise to this to reduce the S/N to 10.
Figure 4 (A
) is the frame power Pi calculated from the voice feature vector Toi in a no-noise environment, and (B) shows the frame power Pi calculated from the voice feature vector Toi in a noisy environment using the same method.
This is the frame power Pi' calculated from . The graph shows time plotted on the horizontal axis and frame power plotted on the vertical axis.

As can be understood from Fig. 4 (A) and (B), the change in the frame power Pi obtained from the speech feature vector Toi that reduces the noise pattern varies depending on the period in which the speech is being produced and the period in which the speech is being produced. It has a clear distinction from the section where there is no. Therefore, voice section detection can be easily performed even in a noisy environment.

[Linear expansion/contraction processing of (“) term] This process is performed by the linear expansion/contraction unit 27, the voice section detection unit 28
The local peak vector between the starting point I5 and the ending point Iε detected by is linearly expanded or contracted on the time axis to a constant audio frame length. The expansion/contraction processing in the linear expansion/contraction section 27 is mainly for facilitating linear matching, which will be described later, and is also a process for facilitating area management when storing comparison patterns, which will be described later, in the memory.

Next, a method of 1-time axis linear expansion/contraction will be explained. Here, for explanation purposes, we will consider the case of linear expansion/contraction to 32 audio frames. The starting end is ■s, the ending is ■, the audio frame number after linear expansion and contraction is i'(i' = 1 to 32),
The audio frame number i before linear expansion and contraction is calculated from the formula, and the local peak vector in the i-th audio frame before linear expansion and contraction is calculated as the local peak vector r, ' in the i'-th audio frame after linear expansion and contraction.
shall be. However, in equation 8), [ ] represents a Gauss symbol.

As a result, the local peak vector sequence 1 s 1 s + 1°'''i'''``E-1''E from the start end to the end is linearly expanded and contracted to become a vector sequence '1 2 '''' i '''' 81 '82. Become.

From now on, unless otherwise specified, we will proceed with the numbering of audio frames after linear expansion and contraction.

[Comparison pattern calculation and storage processing in section (g)] This processing is performed in the comparison pattern storage section 28.

In the specific speaker recognition method that limits speakers, words to be recognized (hereinafter referred to as categories) are uttered in advance,
It is necessary to store in advance a pattern (referred to as a comparison pattern) for expressing the word. Comparison pattern storage section 2
8, such comparison patterns are stored. The method for creating this comparison pattern will be explained below. The process of creating this comparison pattern is called registration process.

Here, for the sake of explanation, the number of categories is assumed to be M.

There is also a method of creating a comparison pattern by uttering the same category several times and taking the average of each pattern, but in this example, a comparison pattern is created for one utterance of the category.

The voice used to create the comparison pattern is called a learning voice.

Now, the m-th learning voice that has been digitized is input to the input terminal 2.
1 to the feature vector calculation unit 22 to calculate the feature vector of the learning speech. On the other hand, the noise pattern calculation unit 23 has previously extracted a noise pattern when no learning speech has been input. Therefore, the speech feature vector calculation section 24 subtracts the noise pattern from the noise subtraction calculation section 23 from the feature vector from the feature vector calculation section 22 to calculate the speech feature vector of the learning speech.

Next, this audio feature vector is changed into a local peak vector in the local peak vector calculation unit 25.

On the other hand, the voice section detection unit 26 calculates the power of the learning voice and detects the start and end points.

Next, the linear expansion/contraction unit 27 performs temporal linear expansion/contraction processing to convert it into a 32-frame-long local peak vector sequence. This local peak vector of the learning speech is especially called the comparison local peak vector, and is called □Sj
It is expressed as

m5j=(ms, l ms, l ”' + ms,
I...r ms, 1 Furthermore, a pattern represented by a vector string of comparison local peak vectors is expressed as Sa, and this is called a comparison pattern.

The comparison pattern Syu for each category name is stored in the comparison pattern storage section 28 together with the corresponding category name C□.

As already explained, the comparison pattern S7. Since the size of I is constant, memory address management when storing a plurality of comparison patterns is extremely easy.

[Linear matching of term (h)] This process is performed by the linear matching section 29.

In contrast to the registration process of creating a comparison pattern as described above, the process of performing a recognition operation is called a recognition process. Therefore, the voice input during recognition processing is referred to as input voice.

The voice section of this input voice is also calculated by the voice section detecting section 26.

In addition, the above-mentioned items (a) to (f) are also applied to the input audio.
A local peak vector lr4 (referred to as an input local peak vector) is obtained by performing the same or similar processing as in the term.

The pattern of input speech expressed by the time series of input local peak vectors from the start to the end in this way is called an input pattern, and is expressed by R.

Furthermore, as already explained, the m-th comparison pattern Sm
is expressed as a time series from the start to the end, and is stored in the comparison pattern storage section 28.

Next, a process for calculating the similarity between the input pattern R and the comparison pattern Sm will be explained.

Methods for calculating the similarity of patterns include non-linear and P matching methods, but in the present invention linear matching is used, which is a simple process.

Input pattern R represented by 32 input local peak vectors Machi and comparison pattern ST represented by 32 comparison local peak vectors -3
The pattern similarity between l and l is defined as: Here, the right-hand subscript t represents the transposition of the vector.

[Determination Processing in Item (i)] This process is performed by the determination processing unit 30, and the maximum value is determined by dividing the degree of pattern similarity found for each category.

Category names Cm and ax corresponding to number m%4g of the comparison pattern giving the maximum value are output from the output terminal 31 as recognition results.

As is clear from the above explanation, in the speech recognition method of the present invention, frame power is calculated using a speech feature vector obtained by removing noise patterns from input speech, and speech section detection is performed. Figure 4 (A) and (
As is clear from the comparison of the frame power η calculated by the voice feature vector and the frame power PI' calculated by the unprocessed feature vector shown in B), there are fewer voice section detection errors. It is possible to recognize input speech with high accuracy even in different environments.

Furthermore, since the pattern similarity calculation process is performed using the local peak vector calculated from the audio feature vector, the calculation process is extremely simple.

In general, since the comparison local peak vector is used for the comparison pattern, the storage capacity thereof can be extremely reduced, and therefore, in addition to the above-mentioned simplification of the arithmetic processing, the speech recognition system can be downsized.

(Example) Hereinafter, an example of the present invention will be described with reference to FIG.

FIG. 5 is a block diagram showing a specific circuit configuration for implementing an embodiment of the speech recognition method of the present invention.

In FIG. 5, 41 is a microphone, 42 is an amplifier for amplifying the audio signal, 43 is a low-pass filter,
44 is an A/D converter that converts audio into a digital signal;
45 is a signal processing processor that calculates a feature vector;
8 is a processor, 47 is a program memory in which a processor program is stored, 48 is a comparison pattern memory for storing comparison patterns, 49 is a working memory, 50
is a noise pattern memory for storing noise patterns, 51
is an interface for outputting recognition results to the outside. However, although in a strict sense an interface circuit is required between each component, this is omitted here.

Stop the bleeding, Eitae! 111 Next, an example of the speech recognition system of the present invention will be explained with reference to FIG.

After input audio from a microphone 41 is amplified by an amplifier 42, a low-pass filter (LPF) 43 removes its low frequency components.

Next, the input audio from which low frequency components have been removed is sampled into 12 bits by the A/D converter 44 at a sampling frequency of, for example, 12 kHz. The processing in the low-pass filter 43 described above is necessary for this sampling, and therefore, for example, a low-pass filter with a cutoff frequency of 5 kHz and an attenuation of 48 dB/oct is used.

The audio digital data sampled by the A/D converter 44 is converted into a feature vector by the signal processor 45. As this signal processing processor 45, for example, 32010 manufactured by TI can be used.

The processor 46 performs processing using the feature vector output from the signal processing processor 45 for each audio frame period, and the contents of the processing are divided into (1) registration processing (2) and recognition processing. Each of these processes will be explained below.

[registration process] This process is divided into the following processes.

Noise pattern calculation process Audio feature vector calculation process Comparison local peak vector calculation process Voice section detection processing Linear expansion/contraction and comparison pattern storage processing Each of these processes will be explained below.

(Noise pattern calculation process) For the registration process, for example, 10 audio frames are determined as a noise section. At this time, the speaker does not speak, and only ambient noise is input from the microphone 41.

This noise input is sent via signal paths (42, 43, 44) to a signal processing processor 45 from which a noise vector is generated which is sequentially stored in working memory 48. When noise vectors for 10 audio frames are stored in this memory 48, these noise vectors are averaged and the average value is stored in the noise pattern memory 50.

(Voice feature vector calculation process) After the end of the noise section, a speech feature vector is calculated by subtracting the noise pattern in the noise pattern memory 50 from the feature vector input from the signal processing processor 45,
This is stored in working memory 48.

Although this processing is performed for each audio frame period, the audio feature vectors until the start point is detected by the audio section detection processing are unnecessary, and therefore, in order to use the working memory 49 effectively, they can be discarded appropriately. To go.

(Comparison local peak vector calculation process) The audio feature vectors stored in the working memory 49 are
Through the processing in section d), it is converted into a comparison local peak vector and stored in the working memory 49. This process is also performed every audio frame period. In addition, comparison local peak vectors before the start edge detection are also discarded as appropriate.

(Voice section detection process) Frame power is calculated from the voice feature vector stored in the working memory 48.

The start and end of audio are determined by comparing this frame power with a threshold value.

(Linear expansion/contraction and comparison pattern storage processing) Among the comparison local peak vectors stored in the working memory 48, the comparison local peak vectors from the start end to the end are temporally linearly expanded/contracted by the process in section (f) and stored in the comparison pattern memory. 48.

The method of creating a comparison pattern using the registration process described below creates one comparison pattern from one training voice, but in order to improve recognition performance, it is necessary to create a comparison pattern from multiple training voices of the same category. It is considered a good idea to create a In this case, a method of averaging the comparison patterns created by multiple utterances into one comparison pattern, a method of having all the comparison patterns for each, and various other methods can be considered, but since the comparison patterns of this invention are not wooden, Detailed explanation will be omitted.

[Recognition processing] This process is further divided into the following processes.

Noise pattern calculation processing Speech feature vector calculation processing Input local peak vector calculation processing Speech section detection processing Linear expansion/contraction processing Pattern similarity calculation processing Judgment processing (L sound pattern calculation processing) Noise situation at the time of registration and at the time of certification Since it is possible that the noise pattern has changed, the noise pattern is calculated again.

It is best to calculate this noise pattern for each song in which words are input, but since the input speed of words is slow or it is easy to vocalize during noise measurement, it is best to calculate the noise pattern by setting a special noise section as appropriate. It would be more realistic to measure the noise pattern in sections.

As in the case of registration, a certain 1O voice frame is defined as a noise section, and the speaker does not make any utterances at this time. In this state, only ambient noise is input from the microphone 41 and sent to the signal processing processor 45 in the same manner as described above.
The resulting noise vectors are sequentially stored in working memory 48. When the noise vectors for one 10￥f voice frame are stored, the average of these noise vectors is taken and this average noise vector is stored in the noise pattern memory 50.

(Speech feature vector extraction process) After the end of the noise section, the speech feature vector is calculated using a new noise pattern.

A voice feature vector is calculated by subtracting the noise pattern stored in the noise pattern memory 50 from the feature vector input from the signal processor 45, and is stored in the working memory 49. This process is performed every audio frame period. Furthermore, since the speech feature vectors before the start edge detection, which will be described later, are unnecessary, they are discarded as appropriate.

(Input local peak vector calculation process) Working memory 4
(d)
By processing the terms, the input local peak vector is converted into an input local peak vector and stored in the working memory 49.

This process is also performed every audio frame period. In addition, input local peak vectors before the start edge detection are also discarded as appropriate.

(Voice section detection process) Frame power Pi is calculated from the voice feature vector stored in the working memory 49. This frame power Pi
The start and end of the audio are determined by comparing the threshold and the threshold.

(Linear expansion/contraction processing) The input local peak vector from the start end to the end end stored in the working memory 48 is temporally linearly expanded/contracted by the process in section (f) and stored in the 1 working memory 49.

(Pattern similarity calculation process) The above-mentioned ( The pattern similarity calculation process (linear matching) in term h) is performed, and as a result, DTn
is stored in the working memory 48.

(Determination Process) Using the pattern similarity measure stored in the working memory 49, perform the determination process in the above-mentioned item (i), and send the category name Cm%a obtained as a result to the interface 51. output to the outside through the

(Effects of the Invention) As is clear from the above description, according to the present invention, the following effects can be obtained.

■Since the frame power is calculated using the speech feature vector from which the noise pattern has been removed from the input and the speech section is detected, there are fewer errors in speech section detection, and therefore the recognition accuracy of the input speech is improved even in noisy environments. Improved than before.

■Since the pattern similarity calculation process is performed using the local peak vector calculated from the audio feature vector,
The arithmetic processing when implementing the speech recognition method of the present invention becomes extremely easy to implement.

■Since the comparison local peak vector is also used for the comparison pattern, its storage capacity is extremely small.Therefore, in addition to the above-mentioned effect (■), the structure of the device for implementing the recognition method of this invention is simple and compact. becomes.

[Brief explanation of drawings]

FIG. 1 is a block diagram for explaining recognition processing of the speech recognition method of the present invention. Figure 2 is a diagram showing the characteristics of the bandpass filter used for speech analysis processing, Figure 3 is an explanatory diagram to explain local peak vector calculation, Figure 4 is a diagram showing the state of frame power, and Figure 5 is FIG. 6 is a block diagram showing an embodiment of the present invention. FIG. 6 is a block diagram for explaining a conventional speech recognition system. 21... Input terminal 22... Feature vector calculation unit 23... Noise pattern calculation unit 24... Voice feature vector calculation unit 25... Local peak vector calculation unit 26... Voice section detection unit 27 ...Linear expansion/contraction section 28...Comparison pattern storage section 28...Linear matching section 30...Judgment section 31...Output terminal 41
... Microphone, 42 ... Amplifier 43 ... Low pass filter, 44 ... A/D converter 45 ... Signal processing processor 46 ... Processor, 47 ... Program memory 48 ... Comparison Pattern memory, 49... Working memory 50... Noise pattern memory 51... Interface. Procedural amendment June 25, 1986

Claims (1)

[Claims] (1) (a) A process of frequency-analyzing input speech and calculating feature vectors, which are vectors of frequency components of the input speech, at fixed time intervals called speech frames, and (b) A predetermined process that is known in advance to be only noise. (c) a process of calculating a voice feature vector by subtracting the noise pattern from the feature vector calculated for each voice frame; (d) For each audio frame, calculate a least squares approximation straight line from the audio feature vector, set the component corresponding to the channel that is maximum in the frequency direction based on the least squares approximation straight line to 1, and set the other components to 1. (e) calculating a frame power in the audio frame from the audio feature vector and detecting the start and end of the audio using the frame power; , (f) a process of linearly expanding and contracting the local peak vector calculated for each audio frame from the start end to the end end to a constant audio frame length, and (g) once or multiple times for each recognition target word in advance. A comparison pattern is calculated by the same or similar processing as the above-mentioned processes (a) to (f) for the learning voice of the utterance, and (h) processing of storing the comparison pattern, and (h) attempting to recognize it. By performing non-linear matching processing between the input pattern obtained by processing the input audio in steps (a) to (f) above and the comparison pattern, the difference between the input pattern and the comparison pattern is determined. It is characterized by comprising a process of calculating pattern similarity, and (i) a process of outputting as a result a category name added to a comparison pattern that gives the maximum pattern similarity among the pattern similarities calculated for each comparison pattern. A voice recognition method that uses